The EUV Monopoly Nobody Can Replicate, Now Trading at a March Discount

ASML builds the machines that build chips. More specifically, it builds lithography systems that use light to etch circuit patterns onto silicon wafers. The more advanced the chip, the shorter the wavelength of light required, and extreme ultraviolet lithography uses a 13.5-nanometer wavelength to produce features so small they are measured in atoms.

The company spent roughly 30 years and several billion euros developing EUV technology. No other company followed. The capital intensity, the complexity of aligning over 100,000 components to sub-nanometer tolerances, and the sheer duration of the development cycle created a moat that competitors never attempted to cross. ASML holds more than 90% of all advanced lithography revenue globally.

The customers are the world's most sophisticated manufacturers: TSMC, Samsung, and Intel are the three primary buyers. Each EUV system costs in excess of $150 million. Delivery lead times run to 18 months or more. The installed base requires continuous service, software upgrades, and spare parts, generating a recurring revenue stream that now accounts for a meaningful share of total sales.

ASML's March 31, 2026 announcement backing European AI data center development signals a broadening of its strategic role beyond chip manufacturing into AI infrastructure supply chains, underscoring that the demand runway extends well beyond cyclical chip cycles.

Revenue compounded from $18.6 billion in 2021 to $32.7 billion in 2025, a 76% increase over four years. Gross margins have held in a tight band between 50.5% and 52.8% across the entire period, with 2025 printing the best gross margin of the five-year span at 52.8%. This is not a company sacrificing margin to buy growth.

Operating margins tell a slightly more textured story. The 36.3% operating margin in 2021 compressed to 30.7% in 2022 as R&D investment in High-NA EUV development accelerated. By 2025 the margin had recovered to 34.6%, the second-best in the period. The 2022 compression was not a sign of structural deterioration. It was the cost of developing the next generation of machines.

Net income reached $9.6 billion in 2025, up 26% from $7.6 billion in 2024. The 2024 figure itself was slightly below 2023's $7.8 billion, a product of timing and product mix rather than demand deterioration. The 2025 number erases any concern about a trend break.

Earnings per share on a trailing basis sits at $28.46. The stock at $1,320 implies a trailing P/E of just under 44 times. For context, this is a company with structural monopoly pricing power, 52% gross margins, and a product backlog measured in years.

The Q4 2025 earnings miss was the most significant in the five-quarter history reviewed here. Actual EPS came in at $7.34 against consensus of $8.60, a 14.7% shortfall. That is a large miss by any standard, and it predictably triggered selling. Chip sector headlines in late March flagged ASML as the dominant story in a broader semiconductor selloff.

But context matters. The prior three quarters in 2024 each beat estimates by 7-10%. Q1 and Q3 of 2025 beat by 3.7% and 1.1% respectively. Q4 2025's miss does not represent a trend in earnings disappointment. It represents a single quarter where delivery timing, customer order sequencing, or mix skewed results below the analyst consensus. Semiconductor equipment revenue is lumpy by design. A machine worth $150 million delivered in Q1 instead of Q4 moves the needle significantly.

Free cash flow, which is less subject to quarterly delivery timing, tells the more honest story. 2025 FCF came in at $10.6 billion, up from $9.5 billion in 2024 and a dramatic recovery from the $3.2 billion trough in 2023. Operating cash flow reached $12.2 billion. These numbers are not consistent with a business that is structurally missing.

The Q4 miss is real. The selloff it helped catalyse created a 14% discount from peak. Whether that discount is a gift or a fair repricing depends almost entirely on your view of the AI-driven semiconductor capex cycle over the next three to five years.

ASML's buyback history reveals a company that manages shareholder returns opportunistically. In 2021, it repurchased $8.6 billion of stock when FCF was at $9.9 billion, an aggressive pace. Buybacks slowed sharply through 2023 and 2024 as FCF dipped and management focused capital on High-NA EUV development.

In 2025, buybacks rebounded to $5.7 billion, running alongside $2.4 billion in dividends for total cash returns of $8.1 billion against $10.6 billion in FCF. That is a 76% cash return ratio. The share count has reduced from 0.42 billion in 2020 to 0.39 billion in 2024, a 7.1% reduction over four years. Modest but consistent.

The balance sheet supports continued aggression: $12.9 billion in cash against only $2.7 billion in total debt. Net cash of roughly $10.2 billion gives the company flexibility to accelerate buybacks or pursue strategic investments without accessing capital markets.

Analysts are pricing in further capital returns. With 22 strong buy ratings, 8 buys, and a consensus target price of $1,461 against the current $1,320, the street sees upside even after the March correction.

The first tailwind is AI compute buildout. Every advanced AI chip, whether an Nvidia GPU, an AMD accelerator, or a custom ASIC from a hyperscaler, requires EUV lithography to manufacture. The capex commitments from TSMC, Samsung, and Intel to expand leading-edge capacity are directly proportional to ASML's order pipeline. The AI investment cycle has not peaked. Data center construction continues to accelerate, and each incremental advanced node fab requires ASML machines.

The second tailwind is the High-NA EUV transition. ASML has commercialized a new generation of machines, High-NA EUV systems, that enable smaller features than first-generation EUV allows. These machines cost in excess of $380 million each, roughly 2.5 times the price of standard EUV. Intel has already taken delivery of early units. TSMC and Samsung will follow. The High-NA transition does not just sustain revenue growth; it resets the revenue per machine upward.

The third tailwind is geopolitical, though with a significant caveat addressed in the risks section. Western semiconductor independence has become a policy priority. The US CHIPS Act, European Chips Act, and equivalent programs in Japan and South Korea are collectively funding new advanced fabs that all require ASML equipment. This government-backed demand is largely insulated from normal cyclical demand patterns and adds a floor under the order book through at least 2028.

The March 31 announcement of ASML backing European AI data center development extends this government-aligned positioning further, signaling that ASML is actively positioning itself as a strategic partner to European technology sovereignty efforts.

Most competitive moats are matters of degree. A brand moat can be eroded by marketing spend. A scale advantage can be closed by a well-funded entrant. ASML's position in EUV lithography is categorically different, and the distinction matters when assessing valuation.

The EUV machine development program began in the 1990s. ASML spent over two decades solving problems that required advances in plasma physics, optics, materials science, and precision engineering simultaneously. The program came close to failing multiple times. No competitor committed to the same path. ASML is now the sole manufacturer of EUV systems not because competitors tried and failed, but because no one else tried.

The switching cost structure is unusual. Chipmakers cannot switch away from ASML EUV without switching away from advanced node manufacturing entirely. There is no alternative supplier, no second-source option, no credible alternative technology on the horizon. The moat is not switching costs in the traditional sense. It is the absence of an exit door.

ASML's installed base of machines generates service revenue that is recurring and sticky. A chip factory running 24 hours a day cannot afford downtime. ASML's service contracts, software subscriptions, and upgrade programs represent a revenue stream that is structurally different from equipment sales. It is less cyclical, higher margin, and grows as the installed base grows.

At 18.7% institutional ownership and 0.8% insider ownership, the stock is not dominated by insider selling pressure. The low institutional ownership figure is surprising for a company of this quality and size, and may reflect the stock's primary European listing and the practical barriers to inclusion in many US-focused mandates.

The trailing P/E of 44 times sounds expensive. The EV/EBITDA of 33.4 times sounds expensive. The price-to-sales of 15 times sounds expensive. Every multiple, in isolation, produces sticker shock.

But multiples are only meaningful relative to growth, durability, and moat quality. ASML grew revenue at roughly 15% per year over the last four years. It posted 52% gross margins. It has no credible competition. Its customers are locked in by the physical constraints of semiconductor manufacturing, not by contracts that can be renegotiated.

For a business with these characteristics, the relevant comparison is not the S&P 500 at 22 times earnings. It is other structural monopolies in critical infrastructure. On that basis, the 44 times trailing multiple, against a forward growth profile driven by AI capex and the High-NA transition, is defensible. Not cheap, but defensible.

The consensus analyst target of $1,461 represents roughly 10.6% upside from the current $1,320 price. That is a modest upside for a stock with 22 strong buy ratings, which suggests analysts are pricing in significant growth but also acknowledging the Q4 earnings overhang.

Sentiment data reinforces a cautiously constructive view. After normalized scores dipping below 0.25 in mid-March amid sector selloff coverage, sentiment has recovered sharply to above 0.94 in the final days of March. That kind of mean-reversion in sentiment, combined with the March 31 piece specifically calling out the monopoly characteristics of the business despite the 14% pullback, suggests the market is beginning to re-engage with the structural thesis rather than the near-term EPS noise.